Yakutian Scientists to Train AI to Locate Mammoth Tusks in Permafrost
In the Republic of Sakha (Yakutia), researchers have begun developing a novel AI-based system to locate mammoth tusks. The Pleistotsen (Pleistocene) project brings together geophysicists, paleontologists and software engineers.
Mammoth ivory extraction in Yakutia has existed for decades, but it has remained largely informal. Prospectors rely on accumulated experience, local terrain cues and chance.
No one knows exactly how many valuable tusks lie beneath the multi-meter layers of permafrost – estimates range from tens of thousands to hundreds of thousands of tons. To date, mammoth ivory has not been formally classified as a mineral resource, preventing its extraction from being fully regulated.
How Pleistotsen Works
The technology combines geophysical survey methods with machine learning. A robotic platform moves across tundra and taiga landscapes, collecting data from multiple sensor types, including ground-penetrating radar, magnetometers and electrical prospecting systems. Each method captures different subsurface characteristics, such as soil density variations, electrical conductivity and magnetic anomalies.
All collected data is fed into a neural network trained on thousands of samples of mammoth bone and surrounding geological material. The algorithm identifies characteristic signatures of bone structures within permafrost layers and outputs coordinates of likely finds with meter-level accuracy. As more discoveries are validated in the field, the system becomes more precise, since the AI continuously learns from both successful detections and errors.
Scientific and Economic Value
The project enables paleontologists to move from isolated discoveries to systematic study of mammoth fauna. Instead of sporadic finds, researchers will be able to analyze entire deposits of remains, reconstruct ancient ecosystems and better understand the causes of megafauna extinction in the region.
For the region, the technology could underpin the development of a legal mammoth ivory market, a material in demand for jewelry and collectors.
Currently, tusk extraction is weakly regulated. Prospectors often operate without licenses, and some finds enter informal circulation. Accurate resource accounting and clear zoning of territories would improve oversight, protect unique paleontological sites and generate additional revenue for the regional budget.
Three Years to a Working System
The project is funded by a grant from the Russian Science Foundation and is designed to run for three years. Researchers say initial practical results could be assessed within two years, when algorithm training is expected to be complete and field tests of the prototype conducted. By the end of the project, developers aim to deliver a fully operational system capable of functioning in Arctic conditions at temperatures as low as minus fifty degrees Celsius.
Special attention is being paid to adapting the equipment to local conditions. The robotic platform must navigate swampy taiga terrain, move across ice layers and operate in strong winds. Engineers are incorporating experience from working in permafrost environments, where temperature fluctuations place additional stress on electronics and mechanical components.
Beyond Tusks
Although the primary goal is to locate mammoth ivory, the technology has broader applications. Combined geophysical and AI methods could be used to identify archaeological sites, assess subsurface ice conditions in the context of climate change and detect risk zones for construction in permafrost regions.
This development illustrates how artificial intelligence is evolving from a tool of the digital economy into an instrument for territorial exploration and management.
